Hi,
The neutron-lib patch is necessary for the use case when the new QoS min_kbps value for the port is 0.
So would be good to have that on Victoria as well.

Regards
Lajos

Slawek Kaplonski <skaplons@redhat.com> ezt írta (időpont: 2020. szept. 16., Sze, 22:21):
Hi,

For me personally it seems ok to merge approve this FFE as this change isn't
very big and is limited only to the QoS service plugin. So IMHO risk of merging
that isn't very big.
There is also scenario test proposed for that feature in [1] so we can ensure
that it is working fine.

On Wed, Sep 16, 2020 at 02:39:27PM +0200, Lajos Katona wrote:
> Hi
> The neutron-lib patch (https://review.opendev.org/750349 ) is a bug fix
> (see [1]) which as do not touch db or API can be backported later in the
> worst case.

Is ther neutron-lib patch necessary to make all of that working so that without
backporting this fix and releasing new version feature in neutron will not work
at all?

> The fix itself doesn't affect other Neutron features, so no harm.
>
> Thanks for your help.
> Regards
> Lajos Katona (lajoskatona)
>
> [1] https://launchpad.net/bugs/1894825
>
> Sean McGinnis <sean.mcginnis@gmx.com> ezt írta (időpont: 2020. szept. 15.,
> K, 18:05):
>
> > > I would like to ask for FFE for the RFE "allow replacing the QoS
> > > policy of bound port", [1].
> > > This feature adds the extra step to port update operation to change
> > > the allocation in Placement to the min_kbps values of the new QoS
> > > policy, if the port has a QoS policy with minimum_bandwidth rule and
> > > is bound and used by a server.
> > >
> > > In neutron there's one open patch:
> > > https://review.opendev.org/747774
> > >
> > > There's an open bug report for the neutron-lib side:
> > > https://bugs.launchpad.net/neutron/+bug/1894825 (placement story:
> > > https://storyboard.openstack.org/#!/story/2008111 )  and a fix for that:
> > > https://review.opendev.org/750349
> > >
> > > [1] https://bugs.launchpad.net/neutron/+bug/1882804
> > >
> > Since this requires an update to neutron-lib, adding [requirements] to
> > the subject. Non-client library freeze was two weeks ago now, so it's a
> > bit late.
> >
> > The fix looks fairly minor, but I don't know that code. Can you comment
> > on the potential risks of this change? We should be stabilizing as much
> > as possible at this point as we approach the final victoria release date.
> >
> > Sean
> >
> >
> >
> >

[1] https://review.opendev.org/#/c/743695

--
Slawek Kaplonski
Senior software engineer
Red Hat